loading . . . Beyond "Rocks versus Clocks": Glimpses of a New Philosophy of Molecular Clock Dating * _This is the latest installment of āProblematica.ā It is written by Max Dresow (gloriously unemployed, but soon to be affiliated with the University of Copenhagen) and Katherine Valde (Loyola University)ā¦_
_Anomalocaris_ reared up, mouth open, its spiny arms outstretched to their furthest limit. It was preparing to fall on a diminutive ālace crabā (_Marrella splendens_), which was paddling like hell to get away. In front of _Marrella_ were plastered the words āEvolutionās Big Bangā in giant, boldface letters. It was all taking place on the cover of _TIME_ magazine (December 4, 1995), whose cover story promised to illuminate how ālife as we know it appeared in an amazing biological frenzy that changed the planet almost overnight.ā
The year 1995 was a remarkable time in Cambrian studies. Long a theater of intense debate about pretty much everything, a fragile consensus had begun to form about the timing and āexplosivenessā of the Cambrian radiation. Life on earth originated around four billion years ago, and for the next three-and-a-half billion years remained mostly confined to the microbial grade. āThen, 543 million years agoā¦, within the span of no more than 10 million years, creatures with teeth and tentacles and claws and jaws materialized with the suddenness of apparitions.ā This was the āCambrian explosion,ā as described by _TIME_ magazineās J. Madeleine Nash, and in 1995, the once-controversial idea was becoming mainstream.
Within a year, however, the āCambrian explosion hypothesisā was again embroiled in controversy. What happened in the interim was the publication of the first āmolecular clockā study to use DNA sequence data to constrain the age of major groups.* This found that āinvertebrates [i.e., protostomes] diverged from chordates [deuterostomes] about a billion years ago, [or] twice as long ago as the Cambrianā (Wray et al. 1996, 568). The date was admittedly not very precise; yet it seemed to refute the contemporary estimateā put forward by James Valentine and colleaguesā that the ancestor of protostomes and deuterostomes lived perhaps thirty million years before the Cambrian, or about 575 million years ago (Valentine et al. 1996).
[* A āmolecular clockā is a method for inferring the age of clades (or, if you prefer, lineage-splitting events), which works on the assumption that molecular sequences change at predictable rates over long timescales.]
Thus (re-) commenced a heated debate in paleontology between advocates of a fairly literal reading of the fossil record and advocates of a long cryptic history for major animal groups. The debate is still unresolved (and so, for that matter, is a parallel debate about the origin of angiosperms, or flowering plants). And yet recentlyā with virtually no hooplaā researchers took a major step towards its resolution. Part of this was technical, and had to do with refinements in molecular clock dating techniques, together with better data. But the more interesting part was philosophical, and concerned a new approach to uncertainty, as well as, perhaps, a new āphilosophy of molecular clock dating.ā
We will attempt to characterize this āphilosophyā in a moment. First, however, it will be useful to fill in some of the history, which explains why the traditional attitude of paleontologists towards molecular clocks has been one of distrust, and even antagonism.
# Molecular clocks and the Cambrian explosion: a history of conflict
The first molecular clock study to investigate the origin of animals was conducted by a paleontologist, Bruce Runnegar. Published in 1982, it used differences in the amino acid sequences of alpha- and beta-hemoglobin proteins to date āthe initial radiation of animal phylaāā as it happens, to about a billion years ago (Runnegar 1982, 14). The study was criticized on methodological grounds, and failed to have much impact on the course of debates about animal origins (e.g., Erwin 1989). However, in 1996, a molecular clock study appeared that āset the paleontological world on its earā (Knoll 2003, 200).
The study emerged from a collaboration between Gregory Wray and Leo Shapiroā both evolutionary biologistsā and Jeffrey Levintonā a paleontologist. It concluded, in line with Runnegarās estimate, that the major animal phyla had emerged between 1.0 and 1.2 billion years ago, or āabout twice as long ago as the Cambrian.ā Unlike Runnegarās study it used DNA sequence data to build the clock, sourced from several independent gene families. Thus, when it concluded that animals had passed nearly half their history without leaving a single unambiguous trace in the rocks, people noticed.
Early molecular clock studies worked on the assumption that molecular sequences change at predictable rates over long timescales. So, given a certain amount of differentiation between the āsameā gene or molecule in different lineages, an investigator could use predetermined rates of sequence change to estimate when the lineages diverged from a common ancestor. Provided that rates of sequence divergence had been properly measured, and provided these rates could be extrapolated across the tree, the method was expected to return at least a reliable ballpark estimate of the true age of a group.
To build the 1996 clock, Wray and colleagues began by gathering nucleotide sequence data for āhomologousā genes in different lineages. Then, they calculated the amount of divergence between pairs of lineages and plotted this against the estimated time of divergence for the two lineages based on the fossil record (see the figure, below). They used vertebrate genes to ācalibrateā the clock, since the vertebrate fossil record was assumed to be pretty good. Anyway, it enabled them to characterize rates of molecular change for each gene that could then be extended to parts of the tree where the fossil record was missing or equivocal.
Sequence divergence rates and estimated interphylum divergence times, from Wray et al. (1996). The reproduced graphic shows two genes, alpha-hemoglobin and NADH 1. Diverge times (in Ma) are plotted on the x-axis; genetic distance is plotted on the y-axis. The data points come from vertebrate taxa with high-quality fossil records. The shaded parts of the plots indicate the full range of āinvertebrate-vertebrate genetic distancesā and the implied range of divergence times
Wray and colleagues were interested in one divergence event in particular: the split between protostomes and deuterostomes. To that end, they measured sequence differences between a large number of protostome and deuterostome species, andā using the rates measured on vertebrate materialsā produced a large number of splitting dates (one per species-pair, per gene). What they found was that āall mean divergence time estimates⦠substantially predate[d] the Cambrian,ā with mean divergence times between protostomes and deuterostomes averaging between 1.0 and 1.2 billion years ago (Wray et al. 1996, 570). The average concealed much variability between the āclocks.ā Genes coding for hemoglobin suggested the split happened about 1.6 billion years ago, for instance, whereas genes coding for cytochrome oxidase suggested a date of just 800 million years. Still, all of these were āincompatible with the Cambrian explosion hypothesis of rapid, shallow interphylum distances.ā Not a year after the _TIME_ magazine cover, notions of explosive evolution in the Cambrian were again on the hot seat.
Many paleontologists were immediately suspicious of the new dates. Asked for a comment, Andy Knoll of Harvard quipped that āmolecular clocks are not Timexes.ā Others shared his caution, and contributed a measure of indignation on behalf of the embattled fossil record (see Benton 2009). The fossil record was incomplete, sureā but could it possibly be _this_ incomplete? For many paleontologists, the idea that animals could have existed for hundreds of millions of years without making a peep was quite literally incredible.
The technical features of Wray and colleaguesā study soon came in for criticism. The study used too few genes to produce reliable estimates, critics said. More seriously, the genes they _did_ use exhibited variability in substitution rate (meaning that some clocks ticked faster than others). Newer analyses, including one published in 1998, revised age-estimates toward conformity with the fossil record (Ayala et al. 1998). But the 1998 study placed the protostome-deuterostome split at about 670 million years ago (Ma), leaving a hundred-million-year gap between the time of the split and the first ambiguous glimmerings of the animal fossil record. Other studies favored older dates, with estimates reduced by perhaps 20ā30% (e.g., Bromham et al. 1998, Wang et al. 1999). So the debate remained polarized.
Around 2008, a group led by Doug Erwin and Kevin Peterson set out to produce the definitive molecular clock study of early animal evolution. To do this, they used ārelaxedā clock methods (designed to accommodate fluctuations in substitution rate) to estimate divergence times for more than a hundred species spread across the tree. What they found was that the last common ancestor of all living animals probably lived around 800 Ma, with the first ācrown groupsā appearing a hundred million years later (Erwin et al. 2011).* The major pulse of bilaterian divergence took place in the Ediacaran Period (then thought to stretch from 635 to about 541 Ma). As for bilaterian crown groups, these appeared between the end of the Ediacaran and the end of the Cambrian, in agreement with the fossil record of well-skeletonized taxa.
[* The ācrown groupā of a clade consists of the last common ancestor of the living members of the clade together with _all_ its descendants, living and extinct. The āstem group,ā by contrast, consists of all the lineages that branched beneath the base of the crown group and above the last common ancestor of the crown group and its closest living relative.]
The conclusion was apparently unavoidable. There had been a lag, perhaps as long as 200 million years, between the origin of animals and their debut in the fossil record. Any account of the Cambrian explosion would have to come to grips with it, or else show that the clock had been improperly wound.
The results of Erwin et al.ās (2011) molecular clock study, showing the estimated divergence times of major clades, superimposed on a colored graphic showing the total number of phyla and classes known from the fossil record of each interval
The result stood for over a decade. Subsequent analyses, using more extensive datasets, only strengthened it, even pushing the origin of bilaterians to a time before the last snowball glaciation (>720 Mya) (Dohrmann and Wƶrheide 2017). It came to seem unlikely that the result was going anywhere: paleontologists were going to have to accept a sizeable gap between the origin of major clades and their first appearance in the fossil record. Yet the most recent molecular clock study of early animal evolution, published in 2024, came to a radically different conclusion (Carlisle et al. 2024). According to this study, animals originated in the early Ediacaran, between 613 and 593 million years ago. Bilaterians originated later, between about 580 and 569 million years ago, and protostomes and deuterostomes later still, between about 575 and 558 million years ago. These are significantly earlier dates than Erwin and company producedā around 200 million years earlier for the origin of animalsā implying missing fossil records of just 20 to 40 million years for major groups.
So, what gives?
# All about uncertainty
To begin, there were some methodological differences between the studies. The Erwin et al. study employed concatenation, for example, treating multiple loci as a single nonrecombining locus (and so, ignoring gene tree discordance) (Tiley et al. 2020). It also employed a more limited dataset, which did not sample the ctenophores, and only sampled three of the four classes of Porifera, or sponges (Dohrmann and Wƶrheide 2017). However, by far the most important difference concerned the handling of fossil evidence, and what came to the same thing, the attitude the investigators took toward _uncertainty_.
Backing up a moment: popular āBayesianā molecular clock methods (employed by Erwin et al. and Carlisle et al.) require the user to specify both minimum and maximum constraints on the ages of branch points, or ānodes.ā* These correspond to the lower and upper limits of admissible ages for those nodes, respectively. Minimum constraints are generally based on the oldest known crown group representative of the clade in question (since the clade cannot be younger than its oldest fossilized member). Maximum constraints, by contrast, āattempt to constrain how much older the clade may be⦠[based] on qualified negative evidence,ā like the evidence of sister taxa in sediments that would be expected to preserve members of the clade in question _had they been present_ (Carlisle et al. 2024, 2). As Carlisle and colleagues observe, āmaxima are often arbitrary but objective, in the sense that they are defined on the basis of a well-studied and readily dated deposit.ā Still, they have often been defined conservatively, indicating that a large envelope of uncertainty surrounds the age of key nodes, which then propagates through the rest of the analysis to produce inflated divergence estimates (relative to the estimates produced by less conservative maxima).
[* For a primer on Bayesian methods of molecular clock dating, see dos Reis et al. (2015).]
Some putative Ediacaran animal fossils from China and South Australia (from Cunningham et al. 2016). Here is the key from the original caption: āA: Embryo-like fossil _Tianzhushania_. B: _Eocyathispongia_. C: _Ramitubus_. D: _Lantiella_. E: _Xiuningella_. F: Putative eumetazoan trace fossil from Mistaken Point, Newfoundland. G: _Helminthoidichnites_ , a putative bilaterian trace fossil. H: _Archaeonassa_ , a putative bilaterian trace fossil.ā
Now consider: there are many reports of animal fossils older than the oldest unambiguous body fossilsā putative sponge spicules, biomarkers, and wiggles in the sediment suggestive of biological activity (e.g., Cunningham et al. 2016). Most of these are probably inorganic productions. And yet there is a non-zero probability that one or the other is genuinely biological, and beyond that, genuinely animal. Would-be clock-makers must ultimately decide what to do about this.
The traditional response has been to adopt conservative maximum constraints, indicating a high degree of uncertainty about the real age of clades. The approach makes sense when the goal of the clock-maker is āto integrate over uncertaintyā (in particular, uncertainties stemming from paleontological disagreements about the interpretation of fossils, and inferences about clade age derived from these interpretations). Yet Carlisle et al. took a different tack: āHere, we seek to minimize this uncertainty, in particular, by departing from the practice of establishing maxima such that they encompass the variable quality of fossil evidence and the broad diversity of interpretations of clade age that are based on itā (Carlisle et al. 2024, 2). The reason was that their goal was _not_ to integrate over uncertainties about the interpretation of the fossil record (producing, in effect, a maximally conservative timescale). It was rather, āto establish an integrative timescale for animal diversification that is based on a critical evaluation of the fossil record and its interpretations rather than an historical consensus viewā (2).
Carlisle et al.ās results
What does it mean to āestablish an integrative timescale for animal diversification⦠based on a critical evaluation of the fossil recordā? In a word: to produce the best possible estimate of the actual ages of clades by excluding low probability interpretations from the analysis. An example of a low probability interpretation is the suggestion that certain sterol molecules found in rocks of Cryogenian age [> 635 Mya] indicate the presence of sponges. This is a low probability interpretation since it is now known that these molecules form āthrough the geological methylation of C29 sterols of chlorophyte algae, the dominant eukaryotes at that timeā (Nettersheim et al. 2019). So, most likely, the presence of putative sponge ābiomarkersā reflects only the presence of algae, not sponges at all.
Could sponges have produced the sterol signature? Yesā but they probably didnāt. So, on the aggressive approach employed by the Carlisle et al. group, this possibility is discounted. It amounts to accepting a trade-off. On the one hand, accommodating the possible truth of low probability interpretations will inflate divergence estimates; on the other, discounting them may produce artificially young estimates if one or more of these interpretations turn out to be true. Neither approach is objectively better than the other. They are simply different things: a best guess, on the one hand, a maximally conservative estimate on the other.
Recognizing this distinction is part of a ānew philosophy of molecular clock dating.ā But it is not the whole thing. Another component of concerns the question of what molecular clocks _are_ , and in a broad sense, what they are _for_.
# Glimpses of a new philosophy of molecular clock dating
To the extent that there was an older āphilosophy of molecular clocks,ā it was organized around the idea that molecular clocks provide an independent line of evidence bearing on the timing of evolutionary events. Of course, everyone knew that molecular clocks were calibrated using fossils: the Wray et al. study, as weāve said, was calibrated using vertebrate genes and fossils, which together permitted a rate of molecular change to be calculated for each gene. But the fossils used for calibration did not bear, evidentially, on the events of interest, which for Wray et al. were about the divergence of basal animal groups. Instead, they were selected because they permitted a rate of molecular change to be calibrated, which could then be extrapolated to deeper parts of the tree.
Newer Bayesian methods have changed this situation. In the specification of minimum constraints, in particular, calibrations are supplied by the very fossil record of the clades under scrutiny. Consequently, there is no question of treating molecules and fossils as independent lines of evidence bearing on the same problem. They are entangled in the very guts of the clock.*
[* The Wray et al. study was truly independent of the fossil record of invertebrates, since nothing about that fossil record constrained the estimates of clade age produced by the clock. Not so Erwin et al. (2011) and Carlisle et al. (2024). Especially the Carlisle et al. study, with its aim of establishing āan integrative timescale for animal diversification⦠based on a critical evaluation of the fossil record,ā hinged on the interpretation of fossil biotas directly relevant to assessing the age of major clades like the metazoans.]
What, then, is a molecular clock? A method for dating the origin of clades, to be sure. But Bayesian clocks are also a means of ā _rationalizing disparate paleontological, molecular, and phylogenetic data, their interpretations, [and] evolutionary models, to obtain a holistic evolutionary timescale_ ā (Carlisle et al. 2024, 9). These newer clocks are not independent lines of evidence bearing on the ages of clades. They are a kind of integrative technology that scaffolds interpretations of the fossil record based on several lines of evidence and a set of coordinating assumptions. As such, they do notā and arguably cannotā conflict with fossil evidence. What molecular clocks (can) conflict with are _interpretations of the fossil record based on fossil evidence_ : in particular, interpretations that āreadā the fossil record literally, or very nearly so. But they do not conflict with fossil evidence as such, because fossil evidence is an integral part of a molecular clock analysis.
None of the preceding remarks is terribly controversial. And yet it is striking that molecular clock methods have not generally been conceptualized as providing a means of interpreting the fossil record. Typically, when they are discussed in relation to the fossil record, the implied relationship is adversarial: it is rocks versus __ clocks, in a battle for all the marbles. And why not? Either animals originated around 550 million years ago (as a literal interpretation of the fossil record might suggest), or they originated >800 million years agoā both of these cannot be true. But this is really a battle over _interpretations_ , which use fossil evidence in different ways, and accord it a different status. It is not ārocks versus clocksā; it is the fossil record alone versus a scaffolded view that combines fossil and molecular evidence into an integrative timescale of early animal evolution. Antagonism is an impoverished idiom for discussing the relationship between molecular and fossil evidence.
# Three Take-Aways
Three more general lessons can be extracted from the preceding discussion. To begin, the new philosophy of molecular clocks highlights the importance, in the historical sciences, of āintegrative technologiesā: methods for rationalizing disparate lines of evidence (together with interpretations and models), which permit investigators to acquire a holistic view of a complex event or phenomenon. Previously, philosophers have stressed the importance of obtaining evidence that ārelies on causally independent processes of trace generation and on conceptually independent detection techniques and inference-warranting bodies of background knowledgeā (Chapman and Wylie 2016, 159). This is surely important. But integration is important too, especially when lines of evidence have potentially complementary strengths and limitations. As a recent paper puts it, āThe fossil record supplies a direct record of past diversity that frequently includes character combinations, ecological associations, and distributions that are not inferable from analyses of recent taxa aloneā (Hopkins et al. 2018, 562). However, āthe molecular record can offer insight on what taxa might be missing from the fossil record, especially for certain time periods and environmentsā (563). The two records thus complement each other: and this means they can be leveraged simultaneously, provided the appropriate scaffolding is in place. This sort of practice warrants more philosophical attention.
A second lesson concerns negative evidence. Although often poo-pooed in philosophical discussions of evidential reasoning, negative evidence features in many historical inferences, to a perhaps surprising degree (Wallach 2019; Brewer 2025). And yet the situation is not as problematic as it seems.
Consider this. In their 2024 study, Carlisle et al. base the maximum constraint on the age of crown group Eumetazoa on the Wengāan biota of South China. This biota, which includes three-dimensionally preserved microfossils, has been interpreted as containing animal embryos. Yet as Carlisle et al. put it, āthe weight of evidence now favors a non-metazoan holozoan affinity [for the embryo-like structures].ā Why then use it as a calibration point? Because the Wengāan location is very likely to have preserved crown group eumetazoans had they been present in the oceans at that time; so the fact that the Wengāan location apparently lacks eumetazoans is strong evidence that eumetazoans had yet to evolve.
Embryo-like fossils from the Wengāan biota (image from Cunningham et al. 2017)
What this shows is that the absence of evidence can be __ quite powerful evidence of absence under the right circumstances. Consider a second example. A recent study has tried to make sense of the absence of a fossil record for early animals by studying the distribution of fossilization processes over time (Anderson et al. 2023). The authors note that most early animals are preserved as carbonaceous remains in fully marine, fine-grained, siliciclastic rocks (āBurgess Shale-type preservationā). However, in the few assemblages characterized by this mode of preservation dated to 789 Ma or earlier, there is no evidence for metazoans.* They conclude that ātaphonomic evidence argues for a maximum constraint of 788.72 ± 0.29 Ma on crown animal [metazoan, not eumetazoan] antiquityā (Anderson et al. 2023, 1067). By characterizing the distribution of fossilization processes over time, and by leveraging background knowledge about the kinds of processes likely to preserve metazoans, scientists can make inference about the likely presence or absence of animal ancestors in fossil biotas.
[* This study is philosophically aligned with the Erwin et al. (2011) study, which sought to provide a conservative estimate for the ages of groups, as opposed to the more aggressive Carlisle et al. (2024) study.]
The third lesson is specific to paleontology. In his 2012 book, _Rereading the Fossil Record_ , David Sepkoski identified three strategies that paleobiologists developed during the 1970s and ā80s to cope with the shortcomings of the fossil record. The first he termed āliteral rereading,ā and involved interpreting the fossil record at (more or less) face value. The second he termed āidealized rereading,ā and involved simulating the history of life _in silico_(basically ignoring the fossil record except for the purpose of comparing its broad patterns to those generated by computer simulations). The third he termed āgeneralized rereading,ā and involved assembling large datasets for the purpose of framing āstatistical generalizations⦠about patterns in lifeās historyā (Sepkoski 2012, 4). Then, after the 1980s, more strategies came online. For example, one of us has previously discussed how paleobiologists began using models of sedimentary accumulation to diagnose and correct for stratigraphic overprint, especially at local and regional scales (Dresow 2023). This, he suggested, was a distinct approach to ārereading the fossil record,ā with particular application to field-based research.
We can now add another approach: the use of molecular clocks to produce integrative timescales for evolutionary events. The approach is limited to Bayesian clock methods, which integrate genomic data with fossil-based time constraints, while also accommodating several independent sources of uncertainty. But it is broad enough to encompass both conservative and aggressive approaches to fossil calibration, which produce different outputs: maximally conservative estimates of clade age versus ābest guessā integrative timescales, respectively.
Again, there is something unfamiliar about regarding molecular clocks as a means of interpreting the fossil record. The trope of ārocks versus clocksā is not easily forgotten, and is reinforced by an apparently irresistible tendency to discuss the relationship in adversarial terms. (Consider the title of a 2016 paper: ācan molecular clocks and the fossil record be reconciled?ā) Yet this is so much historical baggage. Rocks and clocks do not conflict. What conflict are _interpretations_ of the fossil record that either do or do not seek to integrate over multiple sources of information and uncertainty. To use Sepkoskiās language, āliteral rereadingā is at odds with an approach to the fossil record that uses Bayesian methods to obtain integrative evolutionary timescales. But literal rereading is now generally regarded with suspicion; so this tension comes to rather little.
# References
Anderson, R. P., Woltz, C. R., Tosca, N. J., Porter, S. M., and D. E. G. Briggs. 2023. Fossilisation processes and our reading of antiquity. _Trends in Ecology and Evolution_ 38:1060ā1071.
Ayala, F. J., Rzhetsky, A. and F. J. Ayala. 1998. Origin of the metazoan phyla: molecular clocks confirm paleontological estimates. _Proceedings of the National Academy of Science, United States of America_ 95:606ā611.
Benton, M. J. 2009. The fossil record: biological or geological signal? in D. Sepkoski and M. Ruse, eds., _The paleobiological revolution: essays on the growth of modern paleontology._ Chicago: University of Chicago Press, 43ā59.
Brewer, M. forthcoming. āWhen should absence of evidence be evidence of absence? A case study from paleogeology.ā _Philosophy of Science_. doi:10.1017/psa.2025.10125.
Bromham, L., A. Rambaut, R. Fortey, A. Cooper, and D. Penny. 1998. Testing the Cambrian explosion hypothesis using a new molecular dating technique. Proc Natl Acad Sci U S A, 95:12386ā12389.
Carlisle, E., Yin, Z., Pisani, D., and P. C. J. Donoghue, āEdiacaran origin and EdiacaranĀCambrian diversification of Metazoa,ā _Science Advances_ 10 (2024), eadp7161.
Chapman, R. and Wylie, A. 2016. _Evidential reasoning in archaeology_. London: Bloomsbury.
Cunningham, J. A., Liu, A. G., Bengtson, S. and P. C. J. Donoghue. 2016 The origin of animals: can molecular clocks and the fossil record be reconciled? _BioEssays_ 39: 1600120.
Dohrmann, M. and G. Wƶrheide 2017. Dating early animal evolution using phylogenomic data. _Scientific Reports_ 7:3599.
Dos Reis, M., Donoghue, P. C. J., and Z. Yang. 2016. Bayesian molecular clock dating of species divergences in the genomics era. _Nature Reviews Genetics_ 17:71ā80.
Dos Reis, M. Thawornwattana, Y., Angelis, K. Telford, M. J., Donoghue, P. C. J., and Y. Ziheng Yang. 2015. Uncertainty in the timing of the origin of animals and the limits of precision in molecular timescales. _Current Biology_ 25:2939ā2950.
Dresow, M. 2023. Biased, spasmodic and ridiculously incomplete: sequence stratigraphy and the emergence of a new approach to stratigraphic complexity in paleobiology, 1973 _ā_ 1995. _Journal of the History of Biology_ 56:419ā454.
Erwin, D. H. 1989. Molecular clocks, molecular phylogenies and the origin of phyla. _Lethaia_ 22 (1989), 251ā257.
Erwin, D. H., Laflamme, M. Tweedt, S., Sperling, S. A., Pisani, D. and K. Peterson. 2011. Cambrian conundrum: early divergence and later ecological success in the early history of animals. _Science_ 334:1091ā1097.
Hopkins, M., Bapst, D. W., Simpson, C. and R. C. M. Warnock. 2018. The inseparability of sampling and time and its in attempts to unify the molecular and fossil records. _Paleobiology_ 44:561ā574.
Knoll, A. H. 2003. _Life on a Young Planet: The First Three Billion Years of Evolution on Earth._ Princeton: Princeton University Press.
Nash, J. M. 1995.āWhen life exploded. _TIME_ , December 4, 1995.
Nettersheim, B. J. et al. 2019. Putative sponge biomarkers in unicellular Rhizaria question an early rise of animals. _Nature Ecology & Evolution _3:577ā581.
Runnegar, B. 1982. A molecular-clock date for the origin of the animal phyla. _Lethaia_ 15:199ā205.
Sepkoski, D. 2012. _Rereading the fossil record: the growth of paleobiology as an evolutionary discipline._ Chicago: University of Chicago Press.
Tiley, G. P., Poelstra, J. W., dos Reis, M., Yang, Z. and A. D. Yoder. Molecular clocks without rocks: new solutions for old problems. _Trends in Genetics_ 36:845ā856.
Valentine, J. W., Erwin, D. H., and D. Jablonski. 1996. Developmental evolution of metazoan body plans: the fossil evidence,ā _Developmental Biology_ 173:373ā381.
Wallach, E. 2019. Inference from absence: the case of archaeology. _Palgrave Communications_. https://doi.org/10.1057/s41599-019-0307-9.
Wang, D. Y.-C., S. Kumar, and S. B. Hedges. 1999. Divergence time estimates for the early history of animal phyla and the origin of plants, animals, and fungi. _Proceedings of the Royal Society, Part B_ 266:163ā171.
Wray, G. A., Levinton, J. S., and L. H. Shapiro. 1996. Molecular evidence for deep Precambrian divergences among metazoan phyla,ā _Science_ 274:568ā573. http://www.extinctblog.org/extinct/2026/3/14/beyond-rocks-vs-clocks